Method and device for producing highlights of multi-camera sports events based on inter-frame detection

Abstract

The present invention discloses a method and device for producing a multi-camera sports event highlight collection based on inter-frame detection, which belongs to sports event production and includes the steps of: recording live videos by multiple cameras according to the specific scenes of sports events, building a personal identity information database for athletes, performing real-time intelligent analysis on the collected videos, using inter-frame detection, face collection, license plate collection and other means for fusion reasoning, reducing the amount of calculation of ultra-high-definition video by the neural network intelligent editing model in the form of inter-frame picture difference, judging and marking sports wonderful pictures through strip picture segmentation technology, specific athlete face and license plate recognition technology, and deriving the wonderful clips of sports events; and then generating sports highlights. The present invention can quickly and accurately edit sports highlights, and effectively and significantly reduce the amount of neural network calculations, greatly reducing labor, and saving a lot of manpower and machine computing power.

Description

Technical Field

[0001] The present invention relates to the field of sports event production, and more specifically, to a method and device for producing multi-camera sports event highlights based on inter-frame detection. Background Art

[0002] In the production of highlights of previous sports events (such as marathons), post-production editors were often required to use editing tools in the non-editing room to select, edit, and synthesize the materials recorded by multiple cameras at the front end, which wasted a lot of manpower and time. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method and device for producing multi-camera sports event highlights based on inter-frame detection, which can quickly and accurately edit sports highlights and effectively and significantly reduce the amount of neural network calculations, greatly reduce labor, and save a lot of manpower and machine computing power.

[0004] The object of the present invention is achieved through the following solutions:

[0005] A method for producing multi-camera sports event highlights based on inter-frame detection, comprising the steps of:

[0006] After pre-processing in the intelligent editing system, the video server is used to store the live video of the sports event recorded by the live camera of the sports event;

[0007] The fixed sampling and dynamic step length are combined to perform inter-frame judgment on the live video of sports events, and only the pictures with inter-frame difference greater than the set value are processed according to the judgment result;

[0008] For the pictures with inter-frame difference greater than the set value, the sports event video is divided into different camera positions, and the strips are divided according to different camera positions, and then the wonderful moment labels are added;

[0009] Image recognition is used to match highlight moment tags with athlete identities, and highlights are edited using video clips tagged with highlight moments.

[0010] Furthermore, the preprocessing comprises the following sub-steps:

[0011] S1, creating a sports event event in the intelligent editing system, and importing a scene graph of the sports event, determining the positions of the cameras in the scene graph, numbering the cameras in order, and recording;

[0012] S2, inputting the personal identification information of all athletes participating in the event into the intelligent editing system, uploading the facial images of all athletes, and matching and associating each athlete's personal identification information with their facial images.

[0013] Furthermore, the method of performing inter-frame judgment on the live video of the sports event by combining fixed sampling and dynamic step length, and selecting only the pictures whose inter-frame difference is greater than the set value to be processed according to the judgment result, includes the following sub-steps:

[0014] Perform inter-frame judgment processing on the uploaded video, calculate the difference of each frame, and finally save only the video frames whose difference is greater than the set value. When selecting the frame for interpolation detection, calculate at a fixed interval of X frames according to the actual situation, or adaptively adjust the interval of 1-n frames according to the law of picture changes, or combine the two calculation methods, where n is a positive integer.

[0015] Furthermore, the method of dividing the sports event video into different camera positions, dividing the strip images according to the different camera positions, and then labeling the wonderful moments includes the following sub-steps:

[0016] Through pre-processing of the marked camera numbers, the sports event videos are divided into front tracking camera positions, side fixed camera positions, and overhead camera positions. The strip images are divided according to each camera position to judge the moments when athletes surpass, athletes stay, and athletes cross the line, and then the corresponding wonderful moments are labeled.

[0017] Furthermore, after dividing the strip images according to different camera positions, adding wonderful moment labels includes the following sub-steps:

[0018] The cameras used to shoot sports events are divided into three types: ① fixed position on the side of the players, ② following position in front of the players, and ③ overlooking position above the players.

[0019] For situations ① and ②, the strip segmentation is used to decompose the picture vertically into strips of specific pixel sizes. When the characters appear in the same strip, it is determined that there is fierce competition in this frame and marked as a wonderful moment.

[0020] For situation ③, strip segmentation is used to decompose the picture horizontally into strips of specific pixel size. When the characters appear in the same strip, it is determined that there is fierce competition in this frame and it is marked as a wonderful moment.

[0021] Furthermore, the method of matching the wonderful moment tags with the identities of the athletes by using image recognition includes the following sub-steps:

[0022] Find the video clips with personal identification information or facial images corresponding to the athletes in the video, match the identities of the athletes, and edit highlights based on these video clips.

[0023] Furthermore, the method includes the steps of: connecting an APP on a smart phone with the intelligent editing system, allowing a user to query information about an athlete in a sports event through the APP, and using the APP to display personal highlights of the athlete to the user.

[0024] Furthermore, the APP has a paid download function, and users can download athletes' personal highlights by paying.

[0025] Furthermore, the method includes the following steps: applying for cloud live broadcast using a live broadcast platform, configuring the push stream address of the cloud live broadcast to the push stream device of the camera, and configuring the viewing address of the cloud live broadcast to the source location of the video server.

[0026] A computer device comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein when the processor executes the program, any of the above methods is implemented.

[0027] The beneficial effects of the present invention include:

[0028] The method of the embodiment of the present invention is based on the specific scene of the sports event (such as marathon), and multiple cameras are used to collect live videos. At the same time, a personal identification information database of athletes is constructed, and real-time intelligent analysis of the collected videos is performed. Inter-frame detection, face collection, license plate collection and other means are used for fusion reasoning, and the amount of calculation of the neural network intelligent editing model for ultra-high-definition video is reduced by means of inter-frame picture difference. Through strip picture segmentation technology, specific athlete face and license plate recognition technology, the wonderful pictures of sports are judged and marked, and the wonderful clips of sports events are derived; and then sports highlights are generated. The method of the embodiment of the present invention can quickly and accurately edit sports highlights, and effectively and significantly reduce the amount of neural network calculations, greatly reduce labor, and save a lot of manpower and machine computing power.

[0029] The embodiments of the present invention play the role of compressing ultra-high-definition video files, reducing the amount of calculation for identifying highlight nodes, and can accurately locate moments with large picture changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0031] Figure 1 is a flowchart of the steps of the method according to the embodiment of the present invention;

[0032] Figure 2 This is a schematic diagram of the wonderful moments marked when the method of the embodiment of the present invention is for situations ① and ②;

[0033] Figure 3 This is a schematic diagram of a wonderful moment when the method of the embodiment of the present invention is directed to situation ③. DETAILED DESCRIPTION

[0034] All features disclosed in all embodiments in this specification, or steps in all methods or processes implicitly disclosed, except for mutually exclusive features and / or steps, can be combined and / or expanded or replaced in any manner.

[0035] Example 1

[0036] like Figure 1 As shown, a method for producing multi-camera sports event highlights based on inter-frame detection includes the following steps:

[0037] After pre-processing in the intelligent editing system, the video server is used to store the live video of the sports event recorded by the live camera of the sports event;

[0038] The fixed sampling and dynamic step length are combined to perform inter-frame judgment on the live video of sports events, and only the pictures with inter-frame difference greater than the set value are processed according to the judgment result;

[0039] For the pictures with inter-frame difference greater than the set value, the sports event video is divided into different camera positions, and the strips are divided according to different camera positions, and then the wonderful moment labels are added;

[0040] Image recognition is used to match highlight moment tags with athlete identities, and highlights are edited using video clips tagged with highlight moments.

[0041] Example 2

[0042] Based on Example 1, the pretreatment comprises the following sub-steps:

[0043] S1, creating a sports event event in the intelligent editing system, and importing a scene graph of the sports event, determining the positions of the cameras in the scene graph, numbering the cameras in order, and recording;

[0044] S2, inputting the personal identification information of all athletes participating in the event into the intelligent editing system, uploading the facial images of all athletes, and matching and associating each athlete's personal identification information with their facial images.

[0045] In this embodiment, taking a marathon event as an example, the following two methods are used to upload the athlete's personal identification information and facial image, wherein the personal identification information refers to relevant information that can verify the identity of the athlete, specifically including the athlete's name and parameter number.

[0046] Method 1: directly input the athlete's name and competition number on the display front end of the intelligent editing system, and upload the corresponding athlete's face picture to the intelligent editing system for matching and association to ensure that the athlete and his face can correspond.

[0047] Method 2: First, make an Excel spreadsheet with the personal identification information of all participating athletes, and then import the Excel spreadsheet into the system. Then make a separate folder with the facial image of each participating athlete, and name the file in the form of "name + competition number". After uploading, the intelligent editing system will automatically match and associate the corresponding facial image based on the athlete's name and competition number information.

[0048] By associating the athlete's personal identification information with his or her facial image, it can be ensured that the athlete's information is unified in the intelligent editing system, avoiding the problem of mismatch between the athlete and the facial image.

[0049] In this embodiment, the camera is connected to a recording channel, and the recording channel is connected to a recording server. The camera uploads the live video of the sports event to the recording server through the recording channel.

[0050] In the process of collecting live sports event videos, it is necessary to collect the parameters of live sports event videos, including: video length, audio channel, scanning method, video and audio parameters and storage path. In addition, during the collection process, the collection parameters can be modified to collect videos according to actual needs, avoiding secondary processing of the videos later.

[0051] In this embodiment, the recording server stores the live video of the sports event uploaded by the camera in the form of folders. The folders are named according to the numbers of the cameras to facilitate later searches for the live video.

[0052] Example 3

[0053] On the basis of Example 1, the method of combining fixed sampling and dynamic step length to perform inter-frame judgment on the live video of the sports event, and selecting only the pictures whose inter-frame difference is greater than the set value to be processed according to the judgment result, includes the following sub-steps:

[0054] Perform inter-frame judgment processing on the uploaded video, calculate the difference of each frame, and finally save only the video frames whose difference is greater than the set value. When selecting the frame for interpolation detection, calculate at a fixed interval of X frames according to the actual situation, or adaptively adjust the interval of 1-n frames according to the law of picture changes, or combine the two calculation methods, where n is a positive integer.

[0055] Example 4

[0056] On the basis of Example 1, the method of dividing the sports event video into different camera positions, dividing the strip images according to the different camera positions, and then marking the wonderful moment labels includes the following sub-steps:

[0057] Through pre-processing of the marked camera numbers, the sports event videos are divided into front tracking camera positions, side fixed camera positions, and overhead camera positions. The strip images are divided according to each camera position to judge the moments when athletes surpass, athletes stay, and athletes cross the line, and then the corresponding wonderful moments are labeled.

[0058] Example 5

[0059] On the basis of Example 1, after dividing the image strips according to different camera positions, adding wonderful moment labels includes the following sub-steps:

[0060] The cameras used to shoot sports events are divided into three types: ① fixed position on the side of the players, ② following position in front of the players, and ③ overlooking position above the players.

[0061] For situations ① and ②, strip segmentation is used to decompose the picture vertically into strips of specific pixel sizes. When the characters appear in the same strip, it is determined that there is fierce competition in this frame and marked as a wonderful moment. Figure 2 As shown;

[0062] For situation ③, strip segmentation is used to decompose the picture horizontally into strips of specific pixel size. When the characters appear in the same strip, it is judged that there is fierce competition in this frame and marked as a wonderful moment. Figure 3 shown.

[0063] Example 6

[0064] Based on Example 2, the method of matching the wonderful moment tags with the identities of the athletes by using image recognition includes the following sub-steps:

[0065] Find the video clips with personal identification information or facial images corresponding to the athletes in the video, match the identities of the athletes, and edit highlights based on these video clips.

[0066] Example 7

[0067] Based on Example 1, the method includes the following steps: connecting an APP on a smart phone to an intelligent editing system, allowing a user to query information about an athlete in a sports event through the APP, and using the APP to display personal highlights of the athlete to the user.

[0068] Example 8

[0069] Based on Example 7, the APP has a paid download function, and users can download the athlete's personal highlights by paying.

[0070] Example 9

[0071] Based on Example 1, the method includes the following steps: applying for cloud live broadcast using a live broadcast platform, configuring the push stream address of the cloud live broadcast to the push stream device of the camera, and configuring the viewing address of the cloud live broadcast to the source location of the video server.

[0072] Example 10

[0073] A computer device comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein when the processor executes the program, a method as described in any one of Embodiments 1 to 9 is implemented.

[0074] The units involved in the embodiments of the present invention may be implemented by software or hardware, and the units described may also be arranged in a processor. The names of these units do not, in some cases, limit the units themselves.

[0075] According to one aspect of the present application, a computer program product or a computer program is provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in the above-mentioned various optional implementations.

[0076] As another aspect, the present application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiment; or may exist independently without being assembled into the electronic device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by an electronic device, the electronic device implements the method described in the above embodiment.

[0077] The parts not involved in the present invention are the same as the prior art or can be implemented by using the prior art.

[0078] The above technical solution is only one implementation mode of the present invention. For those skilled in the art, it is easy to make various types of improvements or modifications based on the application methods and principles disclosed in the present invention, and it is not limited to the method described in the above specific implementation mode of the present invention. Therefore, the method described above is only preferred and does not have a restrictive meaning.

[0079] In addition to the above examples, those skilled in the art may obtain other embodiments based on the above disclosure or by using the knowledge or technology in the relevant field to make changes. The features of each embodiment may be interchangeable or replaced. The changes and modifications made by those skilled in the art do not depart from the spirit and scope of the present invention and should be within the scope of protection of the claims attached to the present invention.

Claims

1. A method for producing highlights of multi-camera sports events based on inter-frame detection. It is characterized in that Includes steps: After pre-processing in the intelligent editing system, the video server is used to store the live video of the sports event recorded by the live camera of the sports event; The fixed sampling and dynamic step length are combined to perform inter-frame judgment on the live video of sports events, and only the pictures with inter-frame difference greater than the set value are processed according to the judgment result; For the pictures whose inter-frame difference is greater than the set value, the sports event video is divided into different camera positions, and after the strip pictures are divided according to the different camera positions, a wonderful moment label is added; the strip pictures are divided according to the different camera positions, and the wonderful moment label is added, including the sub-steps: The cameras used to shoot sports events are divided into three types: ① fixed position on the side of the players, ② following position in front of the players, and ③ overlooking position above the players. For situations ① and ②, the strip segmentation is used to decompose the picture vertically into strips of specific pixel sizes. When the characters appear in the same strip, it is determined that there is fierce competition in this frame and marked as a wonderful moment. For situation ③, use strip segmentation to decompose the picture horizontally into strips of specific pixel size. When the characters appear in the same strip, it is determined that there is fierce competition in this frame and marked as a wonderful moment. Image recognition is used to match highlight moment tags with athlete identities, and highlights are edited using video clips tagged with highlight moments.

2. The method for producing multi-camera sports event highlights based on inter-frame detection according to claim 1, It is characterized in that The preprocessing comprises the following sub-steps: S1, creating a sports event event in the intelligent editing system, and importing a scene graph of the sports event, determining the positions of the cameras in the scene graph, numbering the cameras in order, and recording; S2, inputting the personal identification information of all athletes participating in the event into the intelligent editing system, uploading the facial images of all athletes, and matching and associating each athlete's personal identification information with their facial images.

3. The method for producing multi-camera sports event highlights based on inter-frame detection according to claim 1, It is characterized in that The method of combining fixed sampling and dynamic step length to perform inter-frame judgment on the live video of the sports event, and selecting to process only the pictures whose inter-frame difference is greater than the set value according to the judgment result, includes the following sub-steps: Perform inter-frame judgment processing on the uploaded video, calculate the difference of each frame, and finally save only the video frames whose difference is greater than the set value. When selecting the frame for interpolation detection, calculate at a fixed interval of X frames according to the actual situation, or adaptively adjust the interval of 1-n frames according to the law of picture changes, or combine the two calculation methods, where n is a positive integer.

4. The method for producing multi-camera sports event highlights based on inter-frame detection according to claim 1, It is characterized in that The method of dividing the sports event video into different camera positions, dividing the strip images according to the different camera positions, and labeling the wonderful moments includes the following sub-steps: Through pre-processing of the marked camera numbers, the sports event videos are divided into front tracking camera positions, side fixed camera positions, and overhead camera positions. The strip images are divided according to each camera position to judge the moments when athletes surpass, athletes stay, and athletes cross the line, and then the corresponding wonderful moments are labeled.

5. The method for producing multi-camera sports event highlights based on inter-frame detection according to claim 2, It is characterized in that The method of matching the wonderful moment tags with the identities of athletes by using image recognition includes the following sub-steps: Find the video clips with personal identification information or facial images corresponding to the athletes in the video, match the identities of the athletes, and edit highlights based on these video clips.

6. The method for producing multi-camera sports event highlights based on inter-frame detection according to claim 1, It is characterized in that The method comprises the following steps: connecting an APP on a smart phone with an intelligent editing system, allowing a user to query information of an athlete in a sports event through the APP, and using the APP to display the athlete's personal highlights to the user.

7. The method for producing multi-camera sports event highlights based on inter-frame detection according to claim 6, It is characterized in that The APP has a paid download function, and users can download athletes’ personal highlights by paying.

8. The method for producing multi-camera sports event highlights based on inter-frame detection according to claim 1, It is characterized in that The method comprises the following steps: applying for cloud live broadcasting by using a live broadcasting platform, configuring the push stream address of the cloud live broadcasting to the push stream device of the camera, and configuring the viewing address of the cloud live broadcasting to the signal source position of the video server.

9. A computer device, It is characterized in that The method comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 8 is implemented.

Images